Not just 64 bit. 64 bit floating point. You can't just get the "low bit". You need to take the exponent into account to figure out which bit of the mantissa to grab. It's intrinsically more computationally difficult than is-odd on a real integer.